Anthropic on Wednesday, 7 October 2026, announced Claude Haiku 5.5, describing it as the company’s “cheapest, fastest, and most capable small model” yet. The new model is designed for high-volume, cost-sensitive workloads and arrives with substantially lower pricing than its predecessor, Haiku 4.5. On average, Anthropic estimates that workloads now cost around 75 percent less to run.
The launch completes the Claude 5.5 family after the earlier releases of Opus 5.5 and Sonnet 5.5, positioning Haiku as the efficient counterpart focused on speed and scale rather than maximum reasoning depth.
Designed for High-Volume and Speed-Sensitive Work
Claude Haiku 5.5 is built to handle quick, repetitive tasks reliably. Anthropic highlights use cases such as document summarisation, content classification, database queries, data compaction and routing. It is also intended to function as a sub-agent alongside the more capable Opus 5.5 and Sonnet 5.5 models, taking on smaller assignments within larger coding or agentic workflows.
Because it is Anthropic’s fastest model to date at standard speeds, Haiku 5.5 is particularly suited to latency-sensitive applications. These include live customer support, real-time browser interactions and in-app assistants where response time matters as much as accuracy.
Significant Price Reductions
The most striking feature of the release is the pricing structure. For prompts up to 100,000 tokens—which Anthropic says account for roughly 90 percent of previous Haiku requests—input tokens are priced at $0.10 per million and output tokens at $0.50 per million. This represents a 90 percent reduction compared with Haiku 4.5.
For longer prompts exceeding 100,000 tokens, the rates rise to $0.50 per million input tokens and $2.50 per million output tokens, still a 50 percent reduction from the previous generation. Cache read and write prices have also been lowered. Anthropic notes that the average 75 percent cost saving takes into account both the new rates and changes in token consumption resulting from an updated tokenizer.
The lower tier pricing matches that of certain competing budget models, placing Haiku 5.5 directly in the competitive landscape of inexpensive, high-throughput options.
Technical Capabilities and Context
Haiku 5.5 supports a 1 million token context window and can generate up to 128,000 output tokens. It accepts text and image inputs and produces text outputs. The model incorporates adaptive thinking with an adjustable effort parameter, a feature that was not available in earlier Haiku versions. The default effort setting is medium.
Anthropic reports meaningful gains over Haiku 4.5 across coding, tool use, computer use and agentic benchmarks. The model is presented as a substantial step forward in capability within the small-model category while retaining the speed and efficiency that define the Haiku line.
Availability Across Platforms
Claude Haiku 5.5 is available immediately on the Claude Platform and through major cloud providers, including Amazon Bedrock, Google Cloud and Microsoft Foundry. It is also accessible to free, Pro, Max, Team and Enterprise users on Claude.ai across web, iOS and Android, and can be used within Claude Code.
Alongside the Haiku launch, Anthropic announced related pricing improvements. Cache reads for Claude Sonnet 5.5 have been halved, making that model approximately 20 percent cheaper for most agentic workloads. The company also introduced a new monthly API credit for Claude Max and Team subscribers to support experimentation with agents and applications.
Positioning in a Competitive Market
The release reflects the intensifying competition in the lower-cost segment of the large language model market. As enterprises and developers seek to run high volumes of routine AI tasks without incurring the expense of frontier models, providers are racing to offer faster and cheaper alternatives. Haiku 5.5 is Anthropic’s answer to that demand, combining improved capability with aggressive price cuts.
By emphasising both performance gains and dramatic cost reductions, Anthropic aims to make sophisticated language model capabilities practical for applications that previously could not justify the expense. The model’s suitability as a sub-agent further encourages architectures in which expensive frontier models handle complex reasoning while smaller, faster models manage the volume of supporting tasks.
Safety and Evaluation
Anthropic stated that Haiku 5.5 has undergone extensive testing against its standards for safety, security and reliability. The company noted that the model includes built-in safeguards for a narrow set of high-risk cybersecurity requests, while most everyday tasks remain unaffected.
A Clear Signal on Value
With Claude Haiku 5.5, Anthropic has sharpened its offering at the efficient end of its model range. The combination of lower prices, higher speed and improved capability is intended to expand the practical use of Claude models in production systems that require both performance and cost control. As organisations continue to integrate AI into high-volume processes, models that deliver reliable results at significantly reduced cost are likely to see rapid adoption.
The launch underscores a broader industry shift: the race is no longer solely about building the most powerful model, but also about delivering capable intelligence at a price and speed that make large-scale deployment economically viable.
Read more – thebnssection.com